glusterd: Get the local txn_info based on trans_id in op_sm call backs.

Issue: when two or more transactions are running concurrently in op_sm,
global op_info might get corrupted.

Fix: Get local txn_info based on trans_id instead of using global txn_info for
commands (re-balance, profile ) which are using op_sm in originator.

TODO: Handle errors properly in call backs and completely remove the global op_info from
op_sm.

Change-Id: I9d61388acc125841ddc77e2bd560cb7f17ae0a5a
BUG: 1229139
Signed-off-by: anand <anekkunt@redhat.com>
Reviewed-on: http://review.gluster.org/11120
Tested-by: Gluster Build System <jenkins@build.gluster.com>
Tested-by: NetBSD Build System <jenkins@build.gluster.org>
Reviewed-by: Krishnan Parthasarathi <kparthas@redhat.com>

5 files changed, 30 insertions, 70 deletions

diff --git a/xlators/mgmt/glusterd/src/glusterd-op-sm.c b/xlators/mgmt/glusterd/src/glusterd-op-sm.c
index bb1ecbb7df3..0a0c6ef0b67 100644
--- a/xlators/mgmt/glusterd/src/glusterd-op-sm.c
+++ b/xlators/mgmt/glusterd/src/glusterd-op-sm.c
@@ -4091,9 +4091,6 @@ glusterd_op_ac_send_stage_op (glusterd_op_sm_event_t *event, void *ctx)
                 goto out;
         }
 
-        if (op == GD_OP_REPLACE_BRICK)
-                glusterd_rb_use_rsp_dict (NULL, rsp_dict);
-
         rcu_read_lock ();
         cds_list_for_each_entry_rcu (peerinfo, &priv->peers, uuid_list) {
                 /* Only send requests to peers who were available before the
diff --git a/xlators/mgmt/glusterd/src/glusterd-rpc-ops.c b/xlators/mgmt/glusterd/src/glusterd-rpc-ops.c
index 00389feb5d8..99f37b05ae9 100644
--- a/xlators/mgmt/glusterd/src/glusterd-rpc-ops.c
+++ b/xlators/mgmt/glusterd/src/glusterd-rpc-ops.c
@@ -1226,11 +1226,6 @@ out:
 
         rcu_read_unlock ();
 
-        switch (rsp.op) {
-        case GD_OP_REPLACE_BRICK:
-                glusterd_rb_use_rsp_dict (NULL, dict);
-                break;
-        }
 
         ret = glusterd_op_sm_inject_event (event_type, txn_id, NULL);
 
@@ -1263,7 +1258,7 @@ int32_t
 __glusterd_commit_op_cbk (struct rpc_req *req, struct iovec *iov,
                           int count, void *myframe)
 {
-        gd1_mgmt_commit_op_rsp         rsp   = {{0},};
+        gd1_mgmt_commit_op_rsp        rsp   = {{0},};
         int                           ret   = -1;
         int32_t                       op_ret = -1;
         glusterd_op_sm_event_type_t   event_type = GD_OP_EVENT_NONE;
@@ -1274,6 +1269,7 @@ __glusterd_commit_op_cbk (struct rpc_req *req, struct iovec *iov,
         xlator_t                      *this = NULL;
         glusterd_conf_t               *priv = NULL;
         uuid_t                        *txn_id = NULL;
+        glusterd_op_info_t            txn_op_info = {{0},};
 
         this = THIS;
         GF_ASSERT (this);
@@ -1343,6 +1339,14 @@ __glusterd_commit_op_cbk (struct rpc_req *req, struct iovec *iov,
         gf_msg_debug (this->name, 0, "transaction ID = %s",
                 uuid_utoa (*txn_id));
 
+        ret = glusterd_get_txn_opinfo (txn_id, &txn_op_info);
+        if (ret) {
+                gf_msg (this->name, GF_LOG_ERROR, 0,
+                        GD_MSG_TRANS_OPINFO_GET_FAIL,
+                        "Failed to get txn_op_info "
+                        "for txn_id = %s", uuid_utoa (*txn_id));
+        }
+
         rcu_read_lock ();
         peerinfo = glusterd_peerinfo_find (rsp.uuid, NULL);
         if (peerinfo == NULL) {
@@ -1372,51 +1376,23 @@ __glusterd_commit_op_cbk (struct rpc_req *req, struct iovec *iov,
                 }
         } else {
                 event_type = GD_OP_EVENT_RCVD_ACC;
-                switch (rsp.op) {
-                case GD_OP_REPLACE_BRICK:
-                        ret = glusterd_rb_use_rsp_dict (NULL, dict);
-                        if (ret)
-                                goto unlock;
-                break;
+                GF_ASSERT (rsp.op == txn_op_info.op);
 
-                case GD_OP_SYNC_VOLUME:
-                        ret = glusterd_sync_use_rsp_dict (NULL, dict);
-                        if (ret)
-                                goto unlock;
-                break;
+                switch (rsp.op) {
 
                 case GD_OP_PROFILE_VOLUME:
-                        ret = glusterd_profile_volume_use_rsp_dict (NULL, dict);
-                        if (ret)
-                                goto unlock;
-                break;
-
-                case GD_OP_GSYNC_SET:
-                        ret = glusterd_gsync_use_rsp_dict (NULL, dict, rsp.op_errstr);
-                        if (ret)
-                                goto unlock;
-                break;
-
-                case GD_OP_STATUS_VOLUME:
-                        ret = glusterd_volume_status_copy_to_op_ctx_dict (NULL, dict);
+                        ret = glusterd_profile_volume_use_rsp_dict (txn_op_info.op_ctx, dict);
                         if (ret)
                                 goto unlock;
                 break;
 
                 case GD_OP_REBALANCE:
                 case GD_OP_DEFRAG_BRICK_VOLUME:
-                        ret = glusterd_volume_rebalance_use_rsp_dict (NULL, dict);
+                        ret = glusterd_volume_rebalance_use_rsp_dict (txn_op_info.op_ctx, dict);
                         if (ret)
                                 goto unlock;
                 break;
 
-                case GD_OP_HEAL_VOLUME:
-                        ret = glusterd_volume_heal_use_rsp_dict (NULL, dict);
-                        if (ret)
-                                goto unlock;
-
-                break;
-
                 default:
                 break;
                 }
diff --git a/xlators/mgmt/glusterd/src/glusterd-syncop.c b/xlators/mgmt/glusterd/src/glusterd-syncop.c
index 9d7b91336ed..1e4d6ce3e2f 100644
--- a/xlators/mgmt/glusterd/src/glusterd-syncop.c
+++ b/xlators/mgmt/glusterd/src/glusterd-syncop.c
@@ -257,12 +257,6 @@ glusterd_syncop_aggr_rsp_dict (glusterd_op_t op, dict_t *aggr, dict_t *rsp)
                         goto out;
         break;
 
-        case GD_OP_PROFILE_VOLUME:
-                ret = glusterd_profile_volume_use_rsp_dict (aggr, rsp);
-                if (ret)
-                        goto out;
-        break;
-
         case GD_OP_GSYNC_CREATE:
         break;
 
@@ -278,12 +272,6 @@ glusterd_syncop_aggr_rsp_dict (glusterd_op_t op, dict_t *aggr, dict_t *rsp)
                         goto out;
         break;
 
-        case GD_OP_REBALANCE:
-        case GD_OP_DEFRAG_BRICK_VOLUME:
-                ret = glusterd_volume_rebalance_use_rsp_dict (aggr, rsp);
-                if (ret)
-                        goto out;
-        break;
 
         case GD_OP_HEAL_VOLUME:
                 ret = glusterd_volume_heal_use_rsp_dict (aggr, rsp);
diff --git a/xlators/mgmt/glusterd/src/glusterd-utils.c b/xlators/mgmt/glusterd/src/glusterd-utils.c
index d21862c76db..05599d178d2 100644
--- a/xlators/mgmt/glusterd/src/glusterd-utils.c
+++ b/xlators/mgmt/glusterd/src/glusterd-utils.c
@@ -7532,23 +7532,26 @@ glusterd_profile_volume_use_rsp_dict (dict_t *aggr, dict_t *rsp_dict)
         int32_t brick_count = 0;
         int32_t count = 0;
         dict_t  *ctx_dict = NULL;
-        glusterd_op_t   op = GD_OP_NONE;
+        xlator_t *this    = NULL;
 
         GF_ASSERT (rsp_dict);
+        this = THIS;
+        GF_ASSERT (this);
 
         ret = dict_get_int32 (rsp_dict, "count", &brick_count);
         if (ret) {
                 ret = 0; //no bricks in the rsp
                 goto out;
         }
-
-        op = glusterd_op_get_op ();
-        GF_ASSERT (GD_OP_PROFILE_VOLUME == op);
         if (aggr) {
                 ctx_dict = aggr;
 
         } else {
-                ctx_dict = glusterd_op_get_ctx ();
+                gf_msg (this->name, GF_LOG_ERROR, 0,
+                        GD_MSG_OPCTX_GET_FAIL,
+                        "Operation Context is not present");
+                ret = -1;
+                goto out;
         }
 
         ret = dict_get_int32 (ctx_dict, "count", &count);
@@ -7971,7 +7974,6 @@ glusterd_volume_rebalance_use_rsp_dict (dict_t *aggr, dict_t *rsp_dict)
         dict_t              *ctx_dict      = NULL;
         double               elapsed_time  = 0;
         glusterd_conf_t     *conf          = NULL;
-        glusterd_op_t        op            = GD_OP_NONE;
         glusterd_peerinfo_t *peerinfo      = NULL;
         glusterd_volinfo_t  *volinfo       = NULL;
         int                  ret           = 0;
@@ -7981,20 +7983,21 @@ glusterd_volume_rebalance_use_rsp_dict (dict_t *aggr, dict_t *rsp_dict)
         int32_t              value32       = 0;
         uint64_t             value         = 0;
         char                *peer_uuid_str = NULL;
+        xlator_t            *this           = NULL;
 
         GF_ASSERT (rsp_dict);
-        conf = THIS->private;
-
-        op = glusterd_op_get_op ();
-        GF_ASSERT ((GD_OP_REBALANCE == op) ||
-                   (GD_OP_DEFRAG_BRICK_VOLUME == op));
+        this = THIS;
+        GF_ASSERT (this);
+        conf = this->private;
 
         if (aggr) {
                 ctx_dict = aggr;
 
         } else {
-                ctx_dict = glusterd_op_get_ctx (op);
-
+                 gf_msg (this->name, GF_LOG_ERROR, 0,
+                         GD_MSG_OPCTX_GET_FAIL,
+                         "Operation Context is not present");
+                goto out;
         }
 
         if (!ctx_dict)
diff --git a/xlators/mgmt/glusterd/src/glusterd-utils.h b/xlators/mgmt/glusterd/src/glusterd-utils.h
index 7bba897629d..84e1fa5d4be 100644
--- a/xlators/mgmt/glusterd/src/glusterd-utils.h
+++ b/xlators/mgmt/glusterd/src/glusterd-utils.h
@@ -427,10 +427,6 @@ int32_t
 glusterd_handle_node_rsp (dict_t *req_ctx, void *pending_entry,
                           glusterd_op_t op, dict_t *rsp_dict, dict_t *op_ctx,
                           char **op_errstr, gd_node_type type);
-int
-glusterd_volume_rebalance_use_rsp_dict (dict_t *aggr, dict_t *rsp_dict);
-int
-glusterd_volume_heal_use_rsp_dict (dict_t *aggr, dict_t *rsp_dict);
 
 int32_t
 glusterd_check_if_quota_trans_enabled (glusterd_volinfo_t *volinfo);